Add support for LoRA #51
Conversation
|
/run-gaudi-tests |
vivekgoe
force-pushed
the
lora
branch
2 times, most recently
from
029856e to
9d92fb2
Compare
|
/run-gaudi-tests |
|
Only codeowners can request to run Gaudi tests. Contact list: kzawora-intel, xuechendi, mswiniarsk, adobrzyn |
vivekgoe
requested review from
kzawora-intel,
xuechendi,
mswiniarsk and
adobrzyn
as code owners
|
Dependent on vllm-project/vllm#21923 merge. |
vivekgoe
force-pushed
the
lora
branch
2 times, most recently
from
a9c2b75 to
385d861
Compare
|
/run-gaudi-tests |
|
@vivekgoe , please add CI trigger in either jenkins folder or tests/full_tests |
|
/run-gaudi-tests |
|
@xuechendi I am not familiar with adding CI triggers, is there any example which you can share which I can follow? |
|
@adobrzyn, how do we suggest for CI, should we add to Jenkins folder - https://github.com/vllm-project/vllm-gaudi/blob/main/.jenkins/test_config.yaml |
|
@xuechendi LoRA related tests in vllm-fork test_config.yaml have not been migrated to vllm-gaudi jenkins config file. Also unit-tests here are triggered in a different way, from here https://github.com/HabanaAI/vllm-fork/blob/42c53f28da01c39fd4eeffbae79f1c8210c41d47/.jenkins/test_config.yaml#L172 . If I understand correctly, any new test added to unit-tests directory will get automatically picked which is happening but lora test is failing on model loading. |
|
Please resolve conficts with main |
|
@adobrzyn @michalkuligowski I will rebase this tomorrow and then will need your help to trigger CI again. |
|
@adobrzyn @michalkuligowski rebased PR, also checked that 2 new unit-tests added for LoRA pass locally. Please retrigger CI and help review the changes. |
|
/run-gaudi-tests |
Remove dependency on LoRA worker class First working version with simple example Fixed BS>1 case Fix in platform.py to avoid error due to missing vllm_config Fix No LoRA case Fix warmup with LoRA Minor Cleanup Disable HPU Graphs Clean-up. Minor fixes Signed-off-by: Vivek <[email protected]> Add LoRA unit-test Signed-off-by: Vivek <[email protected]> Move LoRA configuration code to separate function Signed-off-by: Vivek <[email protected]> Add Multilora test Signed-off-by: Vivek <[email protected]> Fix mypy error Signed-off-by: Vivek <[email protected]> Update hpu_lora to use patching Signed-off-by: Vivek <[email protected]> Fix for model load error in CI Signed-off-by: Vivek <[email protected]>
Signed-off-by: Vivek <[email protected]>
|
/run-gaudi-tests |
Signed-off-by: Vivek <[email protected]>
|
/run-gaudi-tests |
No description provided.